SENTIMENT ANALYSIS GAME MULTIPLAYER ONLINE BATTLE ARENA (MOBA) USING NAÏVE BAYES ALGORITHM Virga Inayatullah Department Statistics, Faculty of Mathematics and Natural Science Islamic University of Indonesia ABSTRACT Multiplayer Online Battle Arena (MOBA) is a subgenre of strategy video games that begin as a subgenre of real time strategy. There are 3 online games based on MOBA that have the most number of downloads and high ratings, namely Mobile Legend, Vainglory, and Arena of Valor. In developing the features of the MOBA game itself, reviews or comments on Google PlayStore are also considered by game developers. Reviews or reviews from MOBA game users generally contain positive suggestions and negative complaints. To sort and monitor these reviews is not an easy thing because the number of reviews published in social media is generally very large when processed manually. There fore we need a method that can sort and monitor these reviews quickly and automatically in categorizing those reviews both positive and negative. The purpose of this study was to classify MOBA game reviews and to extract information from the review. The data used is a review of each game taken from the playstore. Analysis to classify sentiments using the Naive Bayes Classifier. The results from the analysis of the classification of positive and negative sentiments from the game reviews contained in the playstore using the Naive Bayes method, obtained an accuracy of 0.8542 or 85.42%. In addition, from the results of digging up information with text mining, the most common words for the positive sentiment categories are "cool", "like", "fun", "play", and "mobility (Multiplayer Online Battle Arena)". For the category of negative sentiments, the most words are the words "lag" followed by the words "play", "stupid", "bad", "kid", and "afk (Away From Keyboard)". Key Word : Classification, MOBA, Text Mining , NBC